Two Ways Performance Increases Targeting Precision and Response Rates

Performance (specifically, query performance and data load performance in the context of a terabyte-scale data warehouse) increases targeting precision. Better targeting precision increases the relevance of content (or advertising) delivered to an audience through digital media. Increasing relevance leads to all sorts of good things, including:

• Increased campaign response rates for advertisers
• Increased user engagement for publishers (which increases page views and therefore increases advertising inventory)

There are lots of reasons why performance increases targeting precision, but in this post I will focus on two:

1. Audience segment definitions evolve constantly
2. Audience segments are constantly moving through conversion funnels

First, a little background on targeting…advertisers and publishers often target audiences by their demographic and geographic attributes. Contextual targeting is also particularly useful. But behavior is the most powerful indicator of audience intent. Direct response marketers have taken advantage of this fact for decades in order to maximize their campaign response rates.  Behavioral targeting is a hot topic among digital media firms today due to the massive increase in behavioral data now available to marketers, and the power that data delivers for increasing targeting precision.

So where does this data come from? Consumer attention is fragmenting across rapidly proliferating new digital devices (IPTV’s, smartphones, gaming platforms, set top boxes, wireless digital photo frames – it’s a long and expanding list). Advertising budgets are following consumer attention across these new digital channels. And these new digital channels enable the collection of vast quantities of clickstream data (which is – by definition – behavioral data). The presence of this data means that digital channels are direct response channels, and smart marketers leverage direct response techniques in order to increase targeting precision and response rates. In other words, a marketer can deliver a message, measure audience response to that message, and incorporate the results of that measurement into the next message delivered to that audience in order to optimize response.

This IAB presentation speaks to the value of behavioral targeting – it cites data from Advertising.com showing that behavioral targeting increases their response rates somewhere between 300% and 2000%!

Now for the two ways that performance increases targeting precision:

1. Targeting algorithms involving direct response (clickstream) data often require terabyte-scale data analysis – predictive analysis techniques like CHAID, factor analysis, cluster analysis, and logistic regression are often utilized, and as data scale increases, demand for query performance and data load performance increases. And the key point here is that because every audience behavior changes the segment definitions, the ability to recalculate and re-segment the audience in near real-time becomes critical to increasing targeting precision. This exercise is also known as modeling.  Once a predictive model is crafted, it’s possible to reuse the model – applying it to new data in order to make predictions about unseen behavior (known as scoring), but direct response models in the digital media world have increasingly short shelf-lives – so the fresher the model, the more accurate the results, and the higher the response rate.
2. Minimizing the time required for modeling and scoring is important, because each audience segment is associated with a conversion funnel. And the audience is constantly moving through these conversion funnels, so once a segment is defined, the quicker you can reach that segment with your message, the better – because if you wait too long the audience will have advanced in the funnel to a point where your message is no longer relevant to them. For example, you don’t want to waste your money serving somebody an ad for your digital camera if they just bought your competitor’s model an hour ago.

Can you think of additional reasons that query performance and data load performance increase targeting precision and response rates?